Abstract Southern China was a region with mixed rice-millet farming during the Middle Neolithic period and also suggested to be the homeland of Tai-Kadai-speaking (TK) people. The archaeological evidence of animal and plant domestication has demonstrated that southern Chinese rice agriculturalists dispersed from the Yangtze River basin with the dissemination of TK, Austroasiatic (AA), Austronesian (AN) and Hmong-Mien (HM) languages. However, the formations of the inland TK-speaking people, central/southern Han Chinese and their relationships with Neolithic farmers from the Yangtze and Yellow Rivers (YR) basins are far from clear due to the limited sampling of South China. Here, we revealed the spatiotemporal demographic history of southern China by analyzing newly generated genome-wide data of 70 southeastern mainland TK speakers including Dong, Gelao and Bouyei and 45 southwestern Han Chinese together with comprehensive modern/ancient reference datasets. Southwest Han Chinese and Gelao demonstrated a closer genomic affinity to Neolithic YR farmers, while inland TKs (Dong and Bouyei) demonstrated a closer genomic affinity to coastal TK/AN-speaking islanders and Neolithic Yangtze farmers and their descendants. The shared genetic drift between inland TK/AN speaker highlighted a common origin of AN/TK groups, which may be descended from Tanshishan people or their predecessors (Xitoucun). Additionally, we found that inland TK/Sinitic could be modelled as an admixture of ancestral northern East Asian (ANEA) and ancestral southern East Asian (ASEA) sources with different proportions, in which the ANEA was phylogenetically closer to Neolithic millet farmers deriving from the YR Basin and the ASEA was phylogenetically closer to Coastal Neolithic-to-modern southern East Asians. Finally, we discovered genetic differentiation among TK people from southern China and Southeast Asia and obvious substructures between northern and southern inland Chinese TK people. The observed patterns of the spatiotemporal distribution of the northern and southern East Asian lineages in Central/southern China were also compatible with the scenario of bi-directional gene flow events from ANEA and ASEA. Conclusively, multiple lines of genomic evidence indicated millet farmers deriving from the YR basin and rice farmers deriving from the Yangtze River basin substantially contributed to the present-day mainland TK speakers and Central/southern Han Chinese, and formed the modern dual genetic admixture profile.

New insights from the combined discrimination of modern/ancient genome-wide shared alleles and haplotypes: Differentiated demographic history reconstruction of Tai-Kadai and Sinitic people in South China